Low power systems having relatively small currents flowing therein, typically use sleep states in which circuit portions are powered down when not needed to conserve battery charge. Charging capacitors in the circuit up to operating voltage using these small currents requires long periods of time. To overcome the problem of slow power-up, a node to be charged may be brought to power supply voltage through a sufficiently large transistor for an amount of time dictated by a clock. Upon expiration of the appropriate time period, the charging is ceased and the circuit is allowed to settle back to the operating level. The disadvantage of pulling the node to a supply voltage to power-up a circuit is that it has to settle down afterwards. This may take considerable time if the currents available are relatively small. A further disadvantage is that the circuit requires a clock adding additional circuitry and hence consuming additional power.
Another method known to increase power-up speed includes using a kick-start circuit to pump current into a circuit to be powered up. The kick-start circuit provides current to transistors in the circuit being powered up. When the transistors are charged sufficiently, a transistor that produces a logic signal is turned on. The signal then turns the kick-start circuit off, leaving the attached circuitry in a powered-up state.
The disadvantage of a kick-start circuit is that charge pumped into the circuit to be powered up is not related to the amount of charge required to charge the capacitor in the kick-start circuit. Therefore, the kick-start circuit may overshoot the desirable level of charge, and hence, a period of settling down may be necessary.
FIG. 1 depicts a known start-up circuit 100 used in conjunction with a voltage reference circuit 102. Start-up circuit 100 is shown by dotted lines. Voltage reference circuit 102 has two possible equilibrium points, one of which corresponds to zero voltage and zero current, and a second, non-zero equilibrium point, which corresponds to a useful reference voltage. Therefore, voltage reference circuit 102 must be designed to choose only the non-zero equilibrium point to establish the reference voltage. Start-up circuit 100 is provided to allow voltage reference circuit 102 to utilize only the desired equilibrium point. If voltage reference circuit 102 is at the undesired equilibrium point, the voltage is zero and therefore, I.sub.1 and I.sub.2 are zero. Consequently, transistor 104 provides current in transistor 106 which then moves voltage reference circuit 102 to the non-zero equilibrium point. Transistor 104's source voltage increases as the desired equilibrium point is approached. This causes the current through transistor 104 to decrease. When voltage reference circuit 102 reaches the non-zero equilibrium point, the current through transistor 106 will be substantially the same as the current through transistor 108. Transistor 110 and resistor 112 set the gate bias voltage for transistor 104. Voltage reference circuit 102 is on within a gate bias voltage window. Therefore, the gate bias voltage must be high enough to turn voltage reference circuit 102 on but must not exceed the upper limit of the voltage window.
FIG. 2 depicts a kick-start circuit. When current flows in the transistors of the main part of the circuit or band gap reference, the kick-start circuit is turned off. This occurs because MP4 mirrors the current into MN6 which drives the gate of MN3 high and pulls down the drain node of MN3. Driving this node low turns off the current mirrors in the kick-start circuit, so it stops sourcing and sinking current to the band gap reference circuit. R3 ensures that current flows in the kick-start circuit when the band gap reference circuit is powered down.
Conventional circuits do not provide the accuracy and speed desirable to power-up low power systems. Accordingly, there is a need for a start-up circuit that provides a targeted current quickly without significantly overshooting or falling short of the targeted value.